Biomedical subjects

Aidan Winters

Publications and source records attributed to Aidan Winters.

3 recordsLinked to original sources

PubMed2026

Tahoe-100M: Mapping drug-induced molecular phenotypes at single-cell resolution.

We present Tahoe-100M, a giga-scale single-cell perturbation atlas comprising 100 million transcriptomes from 50 diverse cancer cell lines treated with 1,100 drug-dose conditions. This parallel profiling of thousands of perturbations at single-cell resolution with minimal batch effects is enabled by the Mosaic platform, which multiplexes genetically distinct cell models into balanced "cell villages." Beyond cataloging transcriptomic shifts, Tahoe-100M systematically quantifies cellular phenotypes, including proliferation, cytotoxicity, lineage-specific vulnerabilities, and cell-cycle changes. It captures population-level transcriptomic heterogeneity, characterizing whether drug responses drive cells toward divergent fates or convergent states. Pathway-based signatures define drug-induced expression programs, classify mechanisms of action, reveal off-target activities, and expose adaptive stress responses associated with resistance. By unifying cellular and molecular readouts, this broadly applicable perturbation atlas advances our ability to model gene regulation, drug response, and network dynamics. Its public release enables the training of AI frameworks to advance predictive models of cell behavior.

Humans↗

PubMed2026

Predicting cellular responses to perturbation across diverse contexts with State.

While machine learning models offer potential for predicting transcriptomic effects of perturbation, they currently struggle to generalize across cellular contexts. Here, we introduce State, a machine learning model that predicts perturbation effects while accounting for cellular heterogeneity within and across experiments. State is trained using single-cell gene expression data to predict perturbation effects across sets of cells. State improved discrimination of effects on large datasets by more than 30% and identified differentially expressed genes across genetic, signaling, and chemical perturbations with significantly improved accuracy compared with baselines. Its cell embeddings trained on observational data from 167 million cells enable the identification of strong perturbations in cellular contexts where no perturbations were observed during training. We further introduce Cell-Eval, a comprehensive evaluation framework that can be used to evaluate future models. Overall, the performance and flexibility of State set the stage for scaling the development of AI models of cell state.

Machine Learning↗

PubMed2025

Integrated epigenetic and genetic programming of primary human T cells.

Targeted epigenetic engineering of gene expression in cell therapies would allow programming of desirable phenotypes without many of the challenges and safety risks associated with double-strand break-based genetic editing approaches. Here, we develop an all-RNA platform for efficient, durable and multiplexed epigenetic programming in primary human T cells, stably turning endogenous genes off or on using CRISPRoff and CRISPRon epigenetic editors. We achieve epigenetic programming of diverse targeted genomic elements without the need for sustained expression of CRISPR systems. CRISPRoff-mediated gene silencing is maintained through numerous cell divisions, T cell stimulations and in vivo adoptive transfer, avoiding cytotoxicity or chromosomal abnormalities inherent to multiplexed Cas9-mediated genome editing. Lastly, we successfully combined genetic and epigenetic engineering using orthogonal CRISPR Cas12a-dCas9 systems for targeted chimeric antigen receptor (CAR) knock-in and CRISPRoff silencing of therapeutically relevant genes to improve preclinical CAR-T cell-mediated in vivo tumor control and survival.

Humans↗